International Journal of Management, IT & Engineering
Vol. 6 Issue 10, October 2016,
ISSN: 2249-0558 Impact Factor: 6.269
Journal Homepage:, Email: [email protected]
Double-Blind Peer Reviewed Refereed Open Access International Journal - Included in the International Serial
Directories Indexed & Listed at: Ulrich's Periodicals Directory ©, U.S.A., Open J-Gage as well as in Cabell’s
Directories of Publishing Opportunities, U.S.A
J.Suresh babu
The conservation and development of forests are vital to the welfare of human beings. Forests
management is essential to maintain social, economic and ecological services. Forrest
monitoring allows to track their state of health and productivity, in order to conduct proper
management, according to the state of resources, to enhance their functionality and promote
conservation. Remote sensors, optical and radar, offer the possibility of locating changes in
forest areas using various analysis techniques, ranging from the purely visual interpretation to the
implementation of a fully automated algorithm.The main purpose of this work was the
development of a tool to detect in real time (daily) deforestation in the Amazon rainforest, using
satellite images from the MODIS/TERRA sensor and Artificial Neural Networks. The developed
tool provides the parameterization of the configuration for the neural network training to enable
finding the best neural architecture to address the problem and makes use of confusion matrixes
to determine the degree of success of the network. Part of the city of Porto Velho, in Rondônia
state, makes up the tile H11V 09 of the MODIS/TERRA sensor, which was used as the study
area. A spectrum-temporal analysis of this area was made on 57 images from 20 of May to 15 of
July 2003 using the trained neural network. This analysis allowed verifying the quality of the
implemented neural network classification as well as helped the understanding of the dynamics
of deforestation in the Amazon rainforest. The great potential of neural networks for image
classification was perceived with this work. However, the generation of consistent alarms, in
other words, detecting predatory actions at the beginning; instead of firing false alarms is a
complex task that is not yet solved. Therefore, the major contribution of this paper is to provide a
International journal of Management, IT and Engineering, Email: [email protected]
ISSN: 2249-0558 Impact Factor: 6.269
theoretical basis and practical use of neural networks and satellite images to combat illegal
Forests provide support for one billion people that live in extreme poverty around the world, and
provide remunerative employment to more than one hundred million. They contain more than
80% of terrestrial biodiversity (FAO, 2012) and provide essential environmental services such as
soil conservation, watershed management, protection against floods and landslides, and provide
industrial wood (UN). According to the International Tropical Timber Organization (ITTO) of
the UN, it is estimated that deforestation and forest degradation rise 12.9 million hectares per
year and the current area of degraded forest is 850 million hectares. Most of the changes in
forestbased ecosystems due to: a) conversion of land cover, b) land degradation c) intensification
of land use (Lambin, 1997). These changes have resulted in coverage to a wide variety of
ecological impacts, ranging from local to global scale, including changes in productivity and
forests composition, nutrient dynamics, species diversity, and increased atmospheric carbon
dioxide (Braswell et al 2003). One way of assessing changes in land use is based on the
measurement of changes in vegetal and no vegetal cover (Bochco, 2001). Technological progress
allows a comprehensive understanding of any region of the earth's surface from satellite images
(Chuvieco et al, 2002). These images of the Earth have been widely used for change detection,
specifically to the mapping and monitoring deforestation. This has favored international efforts
to establish permanent programs (Vargas and Chuvieco, 1991).
The concepts of Forests and Deforestation were and are defined in different ways depending on
the organization to which we refer. According to FAO (2000) forests and natural plantation with
canopy cover greater than 10% and a surface to 0.5 ha.; determined by the presence of tree and
the absence of other predominant land use, which trees should be able to reach a minimum
height of 5m. In Argentina, the law 26.331/07 of Minimum Standards for Environmental
Protection of Native Forests, defines forests as natural forest ecosystems composed
predominantly of mature native tree spices, with various species of flora and fauna associated,
along with the surrounding medium - soil, subsoil, atmosphere, climate, water-, forming an
interdependent web with its own characteristics and multiple functions, which in its natural state,
International journal of Management, IT and Engineering, Email: [email protected]
ISSN: 2249-0558 Impact Factor: 6.269
give to the system a dynamic equilibrium condition that provides various environmental services
to society, as well as diverse natural resources with the possibility of economic use.
The Program Forest Resources Assessment (FRA, 2006) defines deforestation as the conversion
of forests to other land use or cover reduction, less than 10% of their Total (0.5 ha.). Also the
United Nations Framework Convention on Climate Change (UNFCCC) defines deforestation as
the direct conversion of human-induced forest land to non-forest land.
The lack of a system for the continuous monitoring of the forest cover, including forest inventory
or geographical information system to periodically update the changes in land use, prevent
obtaining actual figures regarding the disappearance of woodlands. To the above, are added,
controls inefficient and little or no supervision over the damage to the forest cover by total and
selective harvesting, authorized annually; therefore, there are no updated figures on the extent of
commercial forest loss and total deforestation (FAO, 1995).
2. Related Works:
Multitemporal analysis techniques from satellite imagery to track deforestation provide a useful
source for the management of the territories affected by this phenomenon. The analysis is based
on the interpretation and comparison of images from different dates and different sensors
(Landsat and Spot). In the case of multi-sensor studies, you should keep in mind that the images
should be compatible. For this, it’s necessary to proceed with the normal pre-processing
(radiometric calibration, geometric and atmospheric correction setting) before starting the
multitemporal analysis. For the study of deforestation is useful to use transformations of images
that reinforce plant component as indexes normalized difference vegetation index (NDVI ) (
critical value for ground cover around 0.1 and 0.5 dense vegetation ) indicating the state of
vitality of vegetation, reducing its value with the loss of vitality and facilitating the identification
of areas of forest cover changes . With a histogram segmentation, forest cover of each image
obtained from NDVI, defined categories of interest (afforestation and reforestation). Are
obtained thus, binary maps of each date and multitemporal crossover is performed to identify
themselves with the changes and / or changes in the vegetation in the area (Chuvieco, 2002
Chuvieco Vargas, 1991).
International journal of Management, IT and Engineering, Email: [email protected]
ISSN: 2249-0558 Impact Factor: 6.269
Another approach is to detect deforestation from "hot spots" on a large spatial scale. "Hot spots"
can be defined as areas that experience a high rate of deforestation in the present and / or in the
recent past. To define different variables can be used, starting with those that allow the initial
identification of large areas susceptible to deforestation, excluding those where the probability of
deforestation in the short term is low and continuing with increasingly relevant variables.
3. Proposed work
3.1 RGB Approach
We will try to detect the change in the vegetation of a particular zone shown in the images over a
time period. We will use the color property i.e. green color for vegetation detection. The
algorithm that we propose for the detection of green area is composed of two parts. In the first
part, the light green shades of the image are detected. In the second part, the dark green
shades are detected.
Detection of Light Green
The RGB value of light green has green component greater than blue and red components. Color
of a pixel in original image, Image1, whose green component is greater than blue and red
components, is copied to a new image, Image2.
In this way, only light green pixels will be copied to Image2. Pixels that do not satisfy the above
property are not copied from Image1 to Image2 and their positions in Image2 are made black.
Detection of Dark Green
The RGB value of dark green has red component of less value than both the green and blue
components. Also, the green component is approximately 20 units less than the blue component.
To identify the dark green pixels, the green component is increased by 20 and the light green
algorithm is applied.
In the above algorithm some shades of grey may also get detected so it may create error in the
So we create an algorithm for prevention of detection for shades of grey.
International journal of Management, IT and Engineering, Email: [email protected]
ISSN: 2249-0558 Impact Factor: 6.269
Prevention of detection for shades of grey.
Intensities of R, G, and B components of green color are approximately less than 100. On the
other hand, intensities of R, G, and B components of grey shades that satisfy also the conditions
of light and dark green are greater than 150. Therefore pixels with R,G and B components
greater than 150 are not considered
HSV stands for hue, saturation, and value, and is also often called HSB (B for brightness). HSV
is cylindrical geometries, with hue, their angular dimension, starting at the red primary at 0°,
passing through the green primary at 120° and the blue primary at 240°, and then wrapping back
to red at 360°. The resulting mixtures in RGB color space can reproduce a wide variety of colors.
In an attempt to accommodate more traditional and intuitive color mixing models, computer
graphics pioneers at PARC and NYIT developed the HSV model in the mid-1970s .
Function available to use in MATLAB library,
hsvImage = rgb2hsv(Image1);
We have used the above function to convert the RGB image to HSV image. converts a redgreen-blue (RGB) to hue-saturation-value (HSV) colormap. Image1 is an m-by-3 matrix, where
m is the number of colors in the colormap. Its columns are intensities of red, green, and blue,
respectively. The columns of hsvImage represent hue, saturation, and value.
4. Result:
Figure 4. Comparison of simulations with 200 samples
International journal of Management, IT and Engineering, Email: [email protected]
ISSN: 2249-0558 Impact Factor: 6.269
5. Conclusion
The conservation and management of forests are very important for the development of human
life. However, this is only possible if conservation measures contemplate economic and social
situations unfair in different countries. Developing countries are those with the largest areas of
tropical forests but also have countless economic and social needs, so that, forest conservation is
not a priority and in many cases, are an obstacle development of economic activities such as
agriculture, oil extraction, extraction of minerals and timber. The forest conservation policies
have to offer financial compensation, to involve communities (especially indigenous peoples),
should be established based on multidisciplinary and jurisdictional studies and must be applied
consistently at all scales, but mostly it is necessary to adopt innovative techniques in order to
make more efficient the use of resources, since the monitoring of forest tracts to measure
deforestation, degradation and / or evolution of the forests are not easy. In this sense, the remote
sensors offer a quick and low cost way to map thousands of acres per day to monitoring
deforestation, logging and other disturbances, as well as, following the recovery of forests. The
access to images is becoming easier and cheaper and analysis techniques range from visual to
those fully automated and can when applied to optical (spectral values) and radar sensors
(backscatter values )
6. References
[1] Lydia P Olander, Holly K Gibbs, Marc Steininger, Jennifer J
Murray , Reference scenarios for
deforestation and forest degradation in support of
a review of data and methods, Scientific Research Centre
of the Slovenian Academy of Sciences and Arts, Ljubljana
(Slovenia), 2001
Krištof Oštir and Tatjana Veljanovski , Application of Satellite remote sensing in Natural
hazard management:
The Mount Mangart Landslide Case Study, Scientific
Centre of the Slovenian Academy of Sciences
Swenson and Brian C
and Arts, Ljubljana (Slovenia), 2002.
Sumalatha Kuthadi, detection of objects from
Department of
high-resolution satellite images,
Computer Science, University of Minnesota Duluth,
Duluth ,
Minnesota 55812 U.S.A , 2004.
International journal of Management, IT and Engineering, Email: [email protected]
ISSN: 2249-0558 Impact Factor: 6.269
GLCF: Data Download Guide [online ]
datadownload.shtml, 2010
copyright © 1994-2011 The MathWorks, Inc [online]
International journal of Management, IT and Engineering, Email: [email protected]